The present invention relates to an error processing apparatus and method of performing error processing for voice-coded data based on ADPCM (Adaptive Differential Pulse Code Modulation).
A simple portable telephone system called a PHS (Personal Handy-phone System), which is in service in recent years, has a transmission frame including unique word information for detecting establishment of synchronization and a CRC (Cyclic Redundancy Check) information for detecting a frame error. However, this system does not have any particular error correction code for voice data transmitted in this frame.
In such a system, when the pull out based on a shift in unique word information is detected on the receiver side, or when a frame error is detected from the CRC information, mute processing of converting the entire error frame into muted data or repeated frame processing of repeatedly using the preceding frame is performed as error processing. However, when mute processing is performed, voice data suddenly becomes muted data, so click noise is generated. In repeated frame processing, voice-coded data after the error is eliminated becomes abnormal.
In the above-described error processing method, noise prevention is insufficient. Accordingly, an error processing method of reducing the difference in ADPCM voice-coded data when the difference is maximum is proposed. Alternatively, an error processing method of suppressing a degradation in voice quality by determining an error generation portion from a frame having an error and converting the corresponding portion is proposed.
More specifically, Japanese Patent Laid-Open No. 4-263528 discloses an error processing method for a digital communication system having a transmission frame including CRC information, in which the transmission frame is divided into a plurality of blocks, a block having an error is determined in the error frame, and only the block which is determined as a block having an error is converted.
FIG. 4 shows the arrangement of the above-described conventional communication system. Referring to FIG. 4, the conventional system comprises a demodulator 20 for demodulating a received frame, an error detection unit 22 for detecting an error in the transmission frame, a first muting circuit 21 for converting ADPCM voice-coded data into muted data depending on the presence/absence of an error, an ADPCM decoder 23 for converting the ADPCM voice-coded data output from the first muting circuit 21 into an analog signal, and a second muting circuit 24 for converting the decoded voice data output from the ADPCM decoder 23 into muted data depending on the presence/absence of an error.
FIG. 5 shows the structure of a transmission frame used in the communication system shown in FIG. 4. Referring to FIG. 5, the transmission frame is constituted by a preamble (PR) 30, a unique word (UW) 31, a voice frame 32 based on ADPCM, and a CRC 33 storing CRC information.
In this communication system, each of the voice frame 32 and the CRC 33 is divided into a plurality of blocks A to D and transmitted. On the reception side, an error block is determined by detecting an error from the CRC information of the CRC 33 in units of blocks A to D.
In the conventional communication system, however, since the error voice-coded data itself cannot be detected, normal voice-coded data in the error frame is rewritten. More specifically, even when the transmission frame is divided into a plurality of blocks to specify the error block, conversion processing is performed for the entire voice-coded data in the error block. For this reason, normal voice-coded data in the error block, which need not be converted, is converted. For this reason, the improvement of voice quality is insufficient in generation of an error frame.
Additionally, in the conventional communication system, in determining an error portion, each of the voice frame 32 and the CRC 33 is divided into a plurality of blocks, and an error block is detected from the CRC information in units of blocks. With this arrangement, the structure of the transmission frame is limited, so the error processing has no versatility.